|
Visualization
Upcoming cosmological observations that survey large patches of the
sky such as the Large Synoptic Survey Telescope project will produce
30 TB per night of data. Simulations carried out to interpret
observations of this type already produce TBs of data per simulation
and this will rise to PBs within a decade. To have any hope of
realizing, encapsulating, and interpreting the enormous wealth of
information contained in such datasets, we have to find very efficient
ways to explore and analyze them, with the goal of eventually
automating many such tasks.
The first step in the analysis process is to actually ``look'' at the
data via a visualization system. Our eye/brain complex still remains
one of the most sensitive analysis tools, albeit (apparently)
optimized for certain types of qualitative analysis. Due to the size
and complexity of the datasets this first step is already highly
nontrivial. For our application, the visualization desiderata consist
of (i) a scalable/interactive system, that (ii) allows hierarchical
(coarse/fine) views of the data, including projections, and (iii) is
steerable in the high-dimensional space of the dataset. The goal is
the seamlessly integrating of the above with quantitative analysis
driven by the visualization process, including the possibility of
on-the-fly definition, implementation, and trials of new analysis
measures.
Code Comparison
|
Code Comparison and Verification: Our
framework for code verification conists of two steps: First
quantitatively identify possible problems with the simulations codes
(left: FLASH has not enough substructure), second postulate a possible
solution and use the visualization tool for quantitative
investigation. In the case shown on the left, the identification of
the "problem" is rather simple: FLASH has much less force
resolution than HOT. But
in very high density regions,
|
| FLASH's resolution is very good (due to
its AMR nature). Therefore, we would expect a similar halo count
in these high density regions for both codes.
The histogram on the right (created with our visualization tool)
shows that this is not the case and that the AMR criterion used
in this case was not stringent enough to capture halos in
the higher density regions. Our new integrated analysis framework
will allow us to identify structures
which can be algorithmically defined such as halos, voids, etc.
These structures can be build up to higher structures (e.g. string of halos
forms filament) and be used as quantitative units to obtain new insights.
|
Code Comparison
|
Publications
-
Multiple Uncertainties in Time-Variant Cosmological Particle
Data,
S. Haroz and K. Heitmann, IEEE Computer Graphics and Applications,
invited paper, accepted for publication, 2008.
-
Provenance in Comparative Analysis: A Study in Cosmology
E. Anderson, C. Silva, J. Ahrens, K. Heitmann, and S. Habib,
Computing in Science and Engineering, in press.
- Multiple Uncertainties in Time-Variant Cosmolgical Particle Data,
Steve Haroz, Kwan-Liu Ma, and Katrin Heitmann, to appear in the Proceedings
of the 2008 Pacific Visualization Symposium,
arXiv:00801.2405
- The Cosmic Code Comparison Project
K. Heitmann, Z. Lukic, P. Fasel, S. Habib, M.S. Warren,
M. White, J. Ahrens, L. Ankeny, R. Armstrong, B.W. O'Shea,
P.M. Ricker, V. Springel, J. Stadel, and H. Trac,
Computations Science and Discovery, invited paper,
arXiv:0706.1270
- Quantitative and Comparative Visualization Applied to Cosmological Simulations,
James Ahrens, Katrin Heitmann, Salman Habib, Lee Ankeny, Patrick McCormick, Jeff Inman, Ryan Armstrong, and Kwan-Liu Ma,
Journal of Physics: Conference Series,
46 (2006)
(article)
|
|
